Research On Fingerprint Ridge Orientation Estimation And Fingerprint Segmentation

Fingerprint is one of the technologies of biometric authentication. It plays a more and more important role in many regions. Although, fingerprint recognition has been extensively studied and many advances have been made, there are still many problems expected to be solved which are shown in actual applications and evaluations (e.g. FVC). As a consequence, in recent years, many academies and industries have been making an in-depth research on fingerprint recognition technologies, so as to improve the performance of fingerprint recognition system and solve the problems in application. This thesis has studied the pre-process of Automatic Fingerprint Identification System (AFIS) on the image level, mainly including ridge orientation estimation and fingerprint segmentation, and quality estimation based on them.There are some features of the fingerprint ridges, which we call gradual change of ridge orientation (GCRO). It shows that: (1) every ridge and furrow change gradually, (2) the difference of orientations between nearby ridges and furrows is small, except the blocks which include the core or the delta point. Our methods are explored based on this feature.1 The criterion of correctness of ridge orientation is proposed based the feature of GCRO, and incorrect orientation are corrected, which improves the correctness of ridge orientation estimation. First, divide the image into W*W blocks, and compute the orientation of each block according to the gradient based method. Then choose a reliable block, from the reliable block to the one next it, sum up the difference of orientation between the current block and 8 blocks around it. The orientation is correct if the difference is less than a threshold given. The wrong orientation is corrected in the end. The algorithm is simple and low computation cost. Better experimental results are gained, comparing to the commonly used algorithm based on gradient and low-pass filter.2 A new fingerprint segmentation method is explored, which include coarse segmentation based on the criterion of correctness of ridge orientation, and amending the coarse segmentation result while correcting wrong orientation. After the orientation field is computed by gradient based method, coarse segmentation is based on the decision of the correctness of its ridge orientation of each block. It is marked as foreground if the orientation is correct or as background if the orientation is not correct. Then the result is amended while correcting incorrect orientation. The blocks wrongly marked as background due to noise is to be remarked as foreground, and the blocks wrongly marked as foreground because of "correct" orientation is to be remarked as background according to the mean and variance of the gray value, which may originally be the blank area without ridges, the totally blurred area with no clear ridges, or the remaining ridges of the previously scanned finger. At last, the entire image is amended with the corrected orientation field. This segmentation method can maintain the area with correct orientation field, and remove the area without ridges, the remaining ridges and the area with incorrect orientation. It helps to extract features of fingerprint better and fast. And it takes no more computation cost because it combines to ridge orientation estimation.